Top-k Multiclass SVM
By Maksim Lapin, Matthias Hein, Bernt Schiele. 2015
97 citations recorded by the source corpus.
Class ambiguity is typical in image classification problems with a large number of classes. When classes are difficult to discriminate, it makes sense to allow k guesses and evaluate classifiers based on the top-k error instead of the standard zero-one loss. We propose top-k multiclass SVM as a direct method to optimize for top-k performance. Our generalization of the well-known multiclass SVM is based on a tight convex upper bound of the top-k error. We propose a fast optimization scheme based on an efficient projection onto the top-k simplex, which is of its own interest. Experiments on five datasets show consistent improvements in top-k accuracy compared to various baselines.
Subfields: Artificial Intelligence, Computational Mechanics, Computer Vision and Pattern Recognition
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
- Bernt Schiele
- Matthias Hein
- Maksim Lapin
- Zeynep Akata
- Trevor Darrell
- Dacheng Tao
- Luc Van Gool
- Gianfranco Doretto
In Computational Mechanics
In Computer Vision and Pattern Recognition
- Matthias Hein
- Bernt Schiele
- Maksim Lapin
- Dacheng Tao
- Luc Van Gool
- Andrew Zisserman
- Trevor Darrell
- Xiaogang Wang
Citations in the corpus
- Loss Functions for Top-k Error: Analysis and Insights (2015, methodology): “Recently, we introduced two top-k versions of the multiclass hinge loss (4) in [23], where the second version is based on the family of ranking losses introduced earlier by [44].”